AI Data Centres and Mental Health: What the Research Actually Shows

Key Takeaways

• This is for anyone living near a proposed or existing data centre, anyone feeling anxious about AI's environmental footprint, and Canadians following the Alberta data centre boom.
• Only one peer-reviewed study directly examines data centres and health impacts (a 2026 exploratory review out of George Mason University). Everything else is built on strong adjacent research applied to a new context.
• The most solid mental health link is financial: a 2025 study in JAMA Network Open found energy-insecure adults had more than double the odds of anxiety and depression compared with energy-secure adults.
• Chronic 24/7 data centre noise (measured at 40 to 59 decibels at some Virginia property lines) is tied to sleep disruption and cardiovascular strain, both established mental health risk factors.
• In Alberta, the Synapse project in Olds and the Wonder Valley project near Grande Prairie have generated real, documented community distress, closer to the psychological toll seen in shale gas and coal mining fights than to any data centre-specific research.

Table of Contents

1. The direct answer
2. What the research actually shows
3. Noise: the clearest physical pathway
4. Financial stress: the strongest mental health link
5. Eco-anxiety and the AI footprint
6. Environmental justice and who gets sited near data centres
7. Community opposition: the Alberta reality
8. Resource diversion and worker mental health
9. What to do if this is affecting you
10. FAQ

Is there a link between AI data centres and mental health?

Yes, but it's an indirect one built mostly on strong adjacent research rather than direct data centre studies. Only one peer-reviewed paper has examined data centres and health outcomes directly: a February 2026 exploratory review in Frontiers in Climate by George Mason University communication researcher Neha Gour and colleagues. Its authors are explicit that the field is new, describing their own work as conceptual and noting the scarcity of peer-reviewed research specific to data centres.

What does hold up is the underlying science each pathway borrows from. Chronic noise exposure is a well-established driver of sleep disruption and cardiovascular strain. Energy and financial insecurity is a well-established driver of anxiety and depression. Contested industrial development near homes is a well-established driver of community psychological distress. AI data centres are now putting real people in the path of all three, particularly across Alberta, where several large projects are working through approval right now.

What the research actually shows: an honest evidence hierarchy

Before going pathway by pathway, it's worth being upfront about evidence quality, since a lot of what circulates online overstates certainty. Three tiers show up repeatedly in this research:

• Peer-reviewed and robust: the noise-health literature (World Health Organization guidelines, epidemiological meta-analyses), the energy insecurity-mental health literature (JAMA Network Open, 2025), and the community-distress literature from contested coal seam gas and shale gas development.
• Emerging or early-stage: the single Frontiers in Climate data centre health review, siting data on environmental justice patterns, and Alberta-specific noise and emissions assessments tied to active project approvals.
• Advocacy framing without direct citation: some of the more alarming figures circulating in social media and campaign materials, including certain infrasound claims and long-range health burden projections, which come from advocacy modelling rather than measured outcomes.

Each section below states which tier applies.

Noise: the clearest physical pathway

Data centres run cooling systems and backup generators around the clock. In Northern Virginia, home to the highest concentration of data centres in North America, measurements at properties that were the subject of noise complaints found levels of 40 to 59 decibels, and some community and industry reporting places typical property-line noise between 55 and 65 decibels. The World Health Organization sets 55 decibels at night as a threshold where adverse health effects begin.

The mental health relevance here comes through an established chain: chronic noise exposure disrupts sleep, raises blood pressure, and increases cardiovascular strain, and all three are recognized risk factors for anxiety and depression. Residents near the Mango Farm data centre complex in Gainesville, Virginia, have described load-testing noise as intense enough to trigger anxiety attacks and prevent sleep entirely. A 2026 urban noise study involving 125 residents found that people living in zones above 65 decibels reported significantly more sleep problems, headaches, and irritability than those in quieter areas.

Evidence tier: the general noise-to-health chain is robust and peer-reviewed. The data centre-specific measurements are real but still limited, drawn mostly from resident complaints and spot checks rather than systematic epidemiological study.

Financial stress: the strongest mental health link

This is the pathway with the most solid mental health evidence behind it, even though the data centre-specific cost question is genuinely contested.

A 2025 study in JAMA Network Open, led by researchers at Georgia Institute of Technology and MetroHealth System/Case Western Reserve University, surveyed adults representing 187 million Americans and found that energy-insecure adults had roughly 2.3 times the odds of both anxiety and depression compared with energy-secure adults. Among people who had to cut spending on necessities just to pay energy bills, close to 4 in 10 reported anxiety symptoms and about a third reported depression symptoms, more than double the rate among those who didn't face that trade-off.

Whether data centres actually push electricity bills up is less settled than headlines suggest. A 2024 Virginia legislative report projected residential bills could rise by $14 to $37 a month by 2040 as data centre demand grows. A 2026 causal analysis using data from the Electric Power Research Institute and Watershed found the opposite over the 2019 to 2024 period, estimating that US retail electricity rates were actually lower than they would have been without data centre-driven investment, while cautioning that future supply constraints could reverse that trend.

In Quebec, Hydro-Québec has proposed a new tariff for data centres over 5 megawatts, roughly double what other large power users currently pay, set to take effect in the second half of 2026. Alberta has taken a different approach: the province largely exempts projects that "bring their own power," meaning developers like Synapse in Olds and O'Leary Ventures at Wonder Valley are building dedicated gas-fired power plants rather than drawing directly from the shared grid. That model would weaken the direct bill-increase pathway for Alberta ratepayers compared to grid-connected US projects, even as it raises separate, well-founded concerns about local air emissions from on-site gas generation.

Evidence tier: the energy insecurity-mental health link is robust and peer-reviewed. Whether data centres raise bills is genuinely disputed in the research and depends heavily on how a given project is powered.

Eco-anxiety and the AI footprint

Eco-anxiety itself is well documented. A 2021 study in The Lancet Planetary Health, surveying 10,000 young people across 10 countries, found that 59 percent were very or extremely worried about climate change, and about 45 percent said those feelings interfered with their daily functioning. A Canadian replication found similar patterns, with at least 58 percent of young Canadians reporting fear, sadness, or powerlessness tied to climate change.

No peer-reviewed study has yet named AI or data centre growth specifically as an eco-anxiety trigger. The concern is a reasonable extension of the existing science, given the scale of water and power some proposed Alberta projects would require, including Wonder Valley's projected 7.5-gigawatt capacity and licensed access to up to 24 million cubic metres of water a year. But the direct link between AI infrastructure specifically and eco-anxiety symptoms remains an inference, not a finding.

Evidence tier: eco-anxiety research is robust; the data centre-specific application is plausible but not yet studied directly.

Environmental justice and who gets sited near data centres

A national US review of roughly 700 data centres found nearly half sit in census tracts with above-median environmental burden, using the CDC's Environmental Justice Index. Other analyses have found a more mixed picture, with residents near some data centres not disproportionately low-income compared to national averages. The pattern likely varies significantly by region and utility structure.

The general link between environmental injustice and mental health disparities is well established in peer-reviewed public health research. In Alberta, the sharpest version of this issue involves Indigenous consultation rather than income. Sturgeon Lake Cree Nation, located roughly 130 kilometres from the proposed Wonder Valley site near Grande Prairie, has challenged a provincial water licence tied to the project, arguing the Crown failed in its duty to consult before the licence was issued. An appeal to Alberta's Environmental Appeals Board was dismissed in April 2026 on the grounds that the nation lacked standing, and a separate judicial review was argued in an Edmonton courtroom in June 2026. Peer-reviewed research on Indigenous land dispossession tied to industrial resource development has found consistent links to negative mental health outcomes and disruption of place-based healing.

Evidence tier: the general environmental justice-mental health link is robust. Data centre-specific siting patterns are still being mapped, and the Alberta situation is better understood through the Indigenous consultation lens than through income-based environmental justice research alone.

Community opposition: the Alberta reality

This is where the evidence, while indirect, maps most closely onto what's actually happening in Alberta right now.

Researchers studying communities affected by coal seam gas and shale gas development have documented real, measurable psychological harm. A 2016 study of 378 Australian farmers found that concern over coal seam gas development was independently associated with depression, anxiety, and stress, with 15 percent falling into a group with clinically significant psychological distress. A 2018 study of nearly 4,800 Pennsylvania patients found that people in areas with the highest shale gas activity were about 1.5 times more likely to report depression symptoms, in a pattern that increased with exposure. Researchers have also documented "solastalgia," the distress caused by watching a familiar home environment change around you, often worsened by a sense of having no control over the outcome.

Alberta's data centre fights fit this pattern closely. Synapse Real Estate Corp. has proposed a CA$10 billion data centre campus in Olds, backed by a 1.4-gigawatt gas-fired power plant located roughly 600 metres from some residents' homes. Residents have described the ongoing hearings as taking over their daily lives, and by June 2026, around 685 people had filed statements of intent to participate with the Alberta Utilities Commission, an unusually high number for this kind of proceeding. The local resident group Olds Residents for Responsible Development has listed noise pollution and mental health impacts explicitly among its stated concerns, alongside worries about air emissions, water use, and falling property values. In March 2026, the AUC rejected Synapse's initial power plant application over what it called significant deficiencies, including an incomplete noise impact assessment; Synapse has since resubmitted, and the process remains ongoing. Farther north, Sturgeon Lake Cree Nation's fight over the Wonder Valley project reflects a related but distinct form of community distress, tied to exclusion from decision-making rather than proximity to construction.

Evidence tier: the underlying research on contested industrial development and community distress is robust and peer-reviewed. No study has yet measured this specifically for data centres, but the Alberta situation is a close real-world match to conditions studied in other industries.

Resource diversion and worker mental health

Two more pathways get raised often but currently rest on weaker ground.

Resource diversion: the argument that massive data centre investment could crowd out funding for mental health services and other public infrastructure appears mainly in opinion and commentary pieces, not in rigorous economic or health policy analysis. It's a reasonable hypothesis, but no study currently quantifies this trade-off directly. This is a genuine research gap rather than an established finding.

Worker mental health: shift work is well studied and clearly linked to worse mental health outcomes. A major meta-analysis covering more than 28,000 workers found shift work raised the risk of adverse mental health outcomes by roughly 28 percent, with an even higher risk for depressive symptoms among women. No research currently looks specifically at data centre workers, so any claim here is an extrapolation from the broader shift-work literature.

Evidence tier: both pathways rely on established adjacent research; neither has direct data centre-specific studies yet.

What to do if this is affecting you

If you're living near a proposed or active data centre and noticing real anxiety, sleep trouble, or stress about your finances or your community, that response is understandable and grounded in real, documented patterns, even in areas where the research is still catching up. A few practical steps:

• Document specifics. Noise logs, sleep disruption, and financial changes give you and any health provider concrete information to work with.
• Connect with others going through it. Community distress research consistently shows that isolation makes psychological strain worse, while organized community groups (like resident associations navigating utility hearings) help people regain a sense of agency.
• Talk to a professional if it's affecting your daily functioning. A therapist can help you process eco-anxiety, financial stress, or the strain of an ongoing land-use fight, without minimizing that the underlying concern is legitimate.

If you're in Alberta and looking for a mental health provider who understands anxiety or the effects of chronic stress, Theralist's directory lets you search by city, specialty, and language across the province, and across Canada.

FAQ

Does living near a data centre definitely cause anxiety or depression?
No single study proves direct causation. The strongest evidence connects data centre-adjacent conditions, like noise and financial strain, to mental health outcomes through well-established general research, not through data centre-specific epidemiology.

What's the single strongest piece of evidence in this whole topic?
A 2025 JAMA Network Open study found energy-insecure adults had roughly 2.3 times the odds of anxiety and depression compared to energy-secure adults. It doesn't study data centres directly, but it's the most rigorous mental health finding that clearly applies to the situation.

Will Alberta data centres raise my power bill?
It depends on the project. Some Alberta developments are pitched as "bring your own power," generating their own electricity rather than drawing on the shared grid, which would weaken the bill-increase risk compared to grid-connected US projects. Hydro-Québec, by contrast, has proposed a new tariff roughly doubling rates for large data centres.

Is this the same as concerns about AI chatbots and mental health?
No. Chatbot research looks at how people use AI tools and their psychological effects. This article covers the physical infrastructure, noise, energy costs, environmental change, and community conflict, that comes with building data centres. They're separate issues.

Where can I read the original data centre health research?
The main peer-reviewed source is Gour, Ortiz, and Maibach's 2026 exploratory review in Frontiers in Climate, examining Northern Virginia. It's the first study of its kind and describes itself as an early, conceptual assessment rather than a definitive one.

Theralist is a Canadian therapist directory built to help people find the right mental health support, wherever they are in Canada. If ongoing stress about your community, finances, or the environment is affecting your daily life, search for a therapist near you who can help.